Data display described: What is it and why it is important

Attracting every hype in these days within the science of the data, but I'm debated at the most important role and often ignored.

When faced with data, there are two important steps:

1. It processes and analyzes information for issuing visual information.
2. To transfer these convenient objects to others.

The second point is important and often neglected. The world's most advanced algorithm or a meaningless understanding is useless if no one can understand. As a data scientist, you should learn to pass your understanding of others. There are more than one reason for this, and a person who is realistic that if good people understand the details, the size of the size will take advantage of. However, there is an equally important reason: it usually describe the findings of others that we find errors, deeply information, or other testing areas.

In this article, we will examine a powerful and effective tool that can help achieve the second step above: Data observation. This is the first time series of articles that will take completely first to the depths of the Realm viewing data. This first article has General and Light, is intended as a representation of the entire field. In the latest articles, I will enter into additional technologies, eventually concluding how you can create your data.

With that information, you will be equipped to deal with your data in new, exciting ways.

“A large number of picture is when it forces us to see what we have never expected to see.” -John Tukey

What is calculated like watching data?

Many people view data recognition for restricted lens, distinguishing normal graphs, such as bar charts, Line charts, and so on, as the perception of true data. Viewed in this vision, DATA visual did not work until the middle of the 18th century. (We will see some examples below.)

However, we do well to grow our minds. A visual conversion of information is not limited to our traditional views. They have been around the thousands of years. For example, here is NAME NAME [1]An old known old map in the world, received as a line of ancient city of Babylon:

This map that puts Babylon at the center and is probably a very useful tool for what we are now identifying geospatial data. He is one of the first data paper on earth.

There is a game of the same mathematical plates. To view these examples as data recognition leads us to an important routine:

In its spine, the data recognition is not just something more than just taking certain data – can be numerous, text, or otherwise – and to use change to declare appearance.

This basic principle leads to several articles related to the most effective ways to carry out these changes, where effective It is freely translated into “faithfulness, easy to understand, and teach. “

The early examples of data recognition

Now that we have increased our views relating to the identity of information, let us look at some modern examples. At the bottom of the chart from 1644 it was built by Michael Florent van Langren [2]. He is one of the old introductions of what we consider traditional mathematical details, indicating a distinction of differences between Rome and Toledo.

Let us look like the next example that follows us – that points exactly the Tukey's quote above.

Below is a London's Soy's District Map in 1854 [3]. Designed by John Snow to determine whether there were a cholera objection patterns that attracted the city at the time:

Looking in the middle of the map, we can see the largest number of death near water pump in the Broad Street. The investigation that determined that the Pump was contaminated and was a major cause of spreading the disease.

This example highlights the goal from John Tukey who has commented above: One of the best ways of data recognition is quickly to see the hardest information to find in the first data form.

Accuracy and flexibility

Data display is a broader topic we can approach in many ways. That means that, two goals to keep in mind regardless of the data viewing user: well doing including Adaptation.

To view good data is not trying to accomplish badly defined activities, such as displaying core of or summarizing Everything IMPORTANT about a set of data. Such statements like these have access to and is actually not possible to achieve.

Instead, good data recognition highlights a particularly specified database aspects that makes it easy to understand the user. You should always specify what you want to show in your personal information before you start designing to see.

Doing within this system, it is helpful to remember that the purpose of viewing data is to start: Display information from the data set and helpful. We want to make data easier to understand. Sure to ensure that we reach this purpose. The visual visit you try to do more can end up confusing viewer even more. It is best to produce a visual insight that includes less data in a clear way. The quality is more important than the value.

See the Data Table below, which contains information about wages from different cities around the United States.

Which of the following is the best choice of the data view above?

1. Mind to simplify information on the data table using the bar chart in Asks in Axts, and uses the color in other cities, and uses lines in the texts (standing lines, etc.) to distinguish between jobs.
2. Similar to see the same, but this without majors. In other words, the word chart and salaries show in the area-based barriers.

It is trying to choose the first, but the truth is, trying to do much. It is better to show limited details, referred to rather than confusing your audience.

In addition to the accuracy, preservation of fluctuations is also important. There is no such thing as the full data view. There is always a place to improve, and data recognition is usually better for each review. Of course, sometimes, data recognition should be shared with others and give its purpose.

This leads to mind – How enough review is? No answer is descriptive for this question. The process of renewal should be done in care. Asking too many people that advice may have a result of a number of cooked, conflicting ideas. On the other hand, publishing first-visual draft – that is, no review at all – may result in the SAPTA Act.

Although there is no perfect solution, there are few guidelines for you to follow:

• Identify 2-3 people to give you the answer in your view.
• Try to make sure your list of people includes the following:
  • An update that is able to edit data view
  • An update with a strong understanding of data used to improve visualization (eg political scientist in election data)
  • An update is part of intended audience in view of eyes
• Pass 2-3 rounds of feedback and review With this same list of people. This will ensure that the development of sighting is continuous and reasonable.

Final thoughts and looking forward

In many ways, DATA view is like writing. Even the coldest and older authors are planned, and their books are traveling through a wide revision before the publication. Why? For a simple reason that good writing depends largely on the audience, and carefully expected reviews ensures the best experience of the visual book of the book. The same idea works in data recognition.

In terms of these guidelines, you can confirm that you are improving the solid data based on the best practices, showing adequate data, and understandable for targeted viewers.

They are the key to accepting data, and the basis for developed programs for the visual views that will be discussed. Until then.

Progress

[1] https://commoned.wisedia.org/wiki/file: The_babylonia_map_a_the_worldhe, _from_sippar _PHROMG
[2] A visual display of the details of the abundanceEdward Tufte
[3]